@Tirth Patel I think you are referring to labeling regions or text when you use the FOTT tool. The tool allows you to select regions that are highlighted in yellow, this is basically done when you select a document from the left side. When OCR runs it identifies text and highlights them, you can then create labels and assign the text to the label so that the model is trained to identify such text while prediction and return these labels. Here is a screen grab of this scenario which could help understand in more detail.
If certain text does not get highlighted after selecting the document you can select Actions-> Run Layout on Current Document to check if such text can still be recognized. In most of the scenarios if the document is of good quality all text including tables are recognized which can then be labeled and trained. Hope this helps!!